Controller Metrics: Close Speed, Accuracy, Control Breaks, Audit Findings

The CFO leans back in the QBR and asks, "How's the accounting function doing?" The Controller says, "Close was clean. No surprises." Heads nod. Slide changes. Everyone moves on.

Three months later, G&A gets a 15% headcount cut. Accounting takes the brunt. The Controller is shocked. They shouldn't be. The scorecard was the problem, not the work.

I've watched this scene play out at half a dozen B2B SaaS companies between $20M and $200M ARR. The Controller is doing real work. The team closes on time. The auditors don't fight. But "clean close" is not a metric. It's a vibe. And vibes get cut first when the CFO trims overhead.

This is the scorecard I'd hand a Controller IC walking into next quarter's QBR. Six metrics. Trended. Benchmarked. Defensible. The CFO can read it in 30 seconds and the Audit Committee can read it in two minutes. That's the whole job.

Why "Clean Close" Is Not A Metric

Look at how every other function shows up to the QBR.

Sales brings cohort retention curves, pipeline coverage, win rates by segment, ramp time by rep. Marketing brings CAC payback, MQL-to-SQL conversion by channel, blended LTV/CAC. Product brings activation funnels, weekly active by feature, churn by cohort. Engineering brings deployment frequency, MTTR, change failure rate.

Accounting shows up with one adjective.

The bar for a credible function isn't "did the work get done." It's trended numbers against a benchmark, with an explanation when the trend bends. Every other team in the company plays at that bar. The Controller has to play there too, or the function reads as overhead.

Three rules for the scorecard:

1. Every metric is trended. Five quarters, minimum. One number in isolation is noise.
2. Every metric has a benchmark. Internal target, peer set, or audit standard. "Down from last quarter" is not enough. The Board wants to know if you're in the right zip code.
3. Every miss has a named remediation play. "We missed the close speed target because rev rec ran long" is fine. "We missed and we're looking into it" is not.

Now the metrics.

Metric 1: Close Speed (Business Days)

Definition: Business days from period close (last day of the month) to "books locked" (final consolidated TB approved, no more JEs accepted).

Formula: business_days_between(period_end_date, books_locked_date)

B2B SaaS benchmark: 5 to 7 business days for a $20M-$200M ARR company on a single ERP. Sub-5 means you're best-in-class or you're cutting corners (more on that in a minute). Above 8 days means you have a sub-process problem (usually rev rec, intercompany, or stock comp).

Why it matters: Close speed is the metric the CFO already cares about. They get asked by the CEO. The CEO gets asked by the Board. If you don't measure it, somebody else is doing it on a napkin.

What to track underneath the headline number:

• Days by sub-process: AP cutoff, AR cutoff, accruals, rev rec, intercompany, equity, consolidations, FX
• Bottleneck of the month (named owner, named cause)
• Trend over five quarters

When I see a close stretching from 6 days to 9 days over three quarters, the cause is almost always rev rec: usage-based revenue, contract amendments, or new product SKUs that didn't get the rev rec policy update. Name it on the slide.

Metric 2: Post-Close Adjustments (Dollars + Count)

Definition: Any journal entry booked after the "books locked" flag, broken into count and dollar value.

Formula: count(JEs where posting_date > books_locked_date AND period = closed_period) and sum(abs(JE_value)) for the same set.

B2B SaaS benchmark: Fewer than 3 adjustments per close, none material (under your audit materiality threshold, usually 1-2% of pre-tax income or revenue, whichever the auditor uses). Zero is the gold standard but rarely realistic.

Why both numbers matter: Many small adjustments and one large adjustment tell completely different stories. Five $2K reclasses means your close process has gaps. Somebody is finding stuff after the fact. One $400K adjustment means somebody made a judgment call that didn't hold up. Different problem, different fix.

When I see eight post-close adjustments in a 4-day close, I know what happened: the team hit the date by deferring work. They locked the books with known issues and cleaned them up after. That's not a fast close. That's a delayed close with a fake stamp on it.

Track:

• Count of post-close JEs
• Total absolute dollar value
• Count above materiality threshold
• Top three accounts by adjustment frequency (these are your weak controls)

Metric 3: Control Breaks

Definition: Any documented control failure: SOX-style if you're a public co or close to it, internal control framework if you're private.

Common controls to track:

• Segregation of duties violations (someone approved their own JE, posted to GL after entering AP)
• Manual JEs above threshold without dual approval
• Bank reconciliation exceptions open more than 30 days
• Wire transfers without dual signoff
• Vendor master changes without independent review
• Revenue contracts above threshold without rev rec sign-off

Target: Zero open at month-end close. Breaks happen. What matters is they get caught, logged, and remediated within the period.

Why it matters more than the audit findings number: Audit findings are lagging. Control breaks are leading. By the time an audit finding shows up, the control has been broken for months. The Controller's job is to find breaks before the auditor does, document them, and close them.

Track:

• Open breaks at period end
• Average age of open breaks
• Breaks by control category
• Breaks logged vs. breaks that became audit findings (the second number should be zero)

A clean control break log with a few self-identified breaks per quarter is better than a log with zero, because zero usually means nobody's looking.

Metric 4: Open Audit Findings (Count + Age)

Definition: Findings from external audit, internal audit, or a SOC 2 audit that are open and not yet remediated.

Tracked by:

• Severity (material weakness, significant deficiency, control deficiency, observation)
• Age bucket (0-30 days, 31-90, 90+)
• Owner (the person on the hook for closing it)

B2B SaaS benchmark:

• Zero open material weaknesses. Period. A material weakness is a fireable offense for a public-co Controller and a major credibility hit for a private one heading toward IPO.
• Fewer than 5 open findings total.
• Zero findings open more than 90 days.

Why aging matters as much as count: A finding that's been open for 6 months says one of two things. Either you don't have a remediation plan, or you have one and it's not working. Both are bad. The Audit Committee will ask about the 90+ day bucket every single time. Be ready.

The owner column is the most important column. A finding without a named owner is a finding nobody's closing. I've seen Controllers run a finding tracker that says "Finance" in the owner column for every line. That's not a tracker. That's a wishlist.

Metric 5: AP/AR Aging Discipline

Definition: Working capital exposures that signal process health, not just liquidity.

Two specific cuts:

AP > 60 days past due (vendor relationship risk). When vendors are unpaid past 60 days, you're either stretching working capital intentionally (CFO call) or you have a broken AP process (Controller call). Either way, the dollar number tells the Board which one it is.

AR > 90 days past due (collectibility risk). At 90+ days, the bad-debt reserve conversation becomes real. The auditors want to see your reserve methodology and your recent collection efforts. If this number is creeping up, your sales team is closing deals that aren't collectible, or your collections process is broken.

Tie to dollar thresholds, not just percentages. "12% of AR is over 90 days" doesn't land. "$1.4M of AR is over 90 days, up from $600K two quarters ago" lands.

B2B SaaS benchmark for AR aging:

• Under 5% of AR over 90 days past due is healthy
• 5-10% is a yellow flag
• Over 10% means you have a collections problem, a billing problem, or a sales-quality problem

For AP:

• Under 2% over 60 days past due is healthy
• Over 5% means either deliberate stretching (which the CFO should know about) or a process break

Metric 6: Accrual Accuracy %

Definition: Variance between the prior-month accrual booked and the actual amount that landed.

Formula: (actual - accrual) / accrual per accrual, then aggregate.

B2B SaaS benchmark:

• Within 5% on the aggregate accrual book
• No single material accrual off by more than 15%

Why this is the metric Controllers under-report: Most Controllers don't track this. They book the accrual, the actual lands, the variance hits the P&L, and nobody compares. Six months later, the FP&A team is wondering why the forecast keeps missing, because the accruals are guesses, not estimates.

The simple test: Is the accrual a real number based on a process, or is it last month's number plus a vibe? When accrual variance runs above 10% consistently, the team is guessing. Time for a process audit.

Where this metric earns its keep: Audit. When the external auditor asks how you arrived at the bonus accrual, the legal accrual, the rev rec contra, or the deferred commission, you can show a five-quarter accuracy track record. That's the difference between a one-meeting conversation and a three-meeting walkthrough.

The "Fast Close But High Adjustments" Diagnostic

Here's the diagnostic that catches more Controllers than any other.

Plot every monthly close on a 2x2:

                     POST-CLOSE ADJUSTMENTS
                  Low (<3)        High (>5)
                ┌──────────────┬──────────────┐
        Fast    │  Best in     │  FAST AND    │
        (<6 d)  │  class       │  SLOPPY      │
                │              │  (danger)    │
                ├──────────────┼──────────────┤
CLOSE   Slow    │  Slow but    │  Slow and    │
SPEED   (>7 d)  │  thorough    │  sloppy      │
                │              │  (rare)      │
                └──────────────┴──────────────┘

A 4-day close with 8 post-close adjustments is worse than a 7-day close with zero. The first one is the team hitting the date by deferring work. The second one is the team doing the job and taking the time it takes.

The bottom-right quadrant ("slow and sloppy") almost never exists in real life. If you're slow, you usually catch the issues before you lock. The dangerous quadrant is top-right: fast and sloppy. That's where Controllers get fired, because the CEO and CFO see the speed number and assume quality. Then the auditor finds the accumulated mess and the conversation gets ugly.

When I see a Controller proud of a 4-day close, my first question is "how many post-close adjustments?" If the answer is "I don't track that," I know what's coming.

The QBR Slide

One slide. Six metrics. Five quarters of trend. Benchmark. RAG (red/amber/green) status. The CFO reads it in 30 seconds.

Here's the layout, copy-pastable, fits on one slide in standard 16:9:

A few notes on building the slide for real:

• Always show benchmark on the row. Don't make the reader remember it from a prior slide.
• Color the cell, not the row. A Controller with five greens and one red on Q-1 is a Controller with a credible function and one named issue. That reads better than the whole row going red.
• Footnote the misses. "Q-2 close speed of 8.5 days driven by rev rec policy update following ASC 606 amendment (see slide 12)." One line. The CFO knows you know.
• Bring the trend, not just the current quarter. The current number is half the story; the slope is the other half.

What To Do When You're Off Benchmark

A scorecard without remediation plays is just a complaint. Here's what to run when each metric goes red.

Close speed > 8 business days

Play: Subprocess audit on the slowest cell. Map every task in the close, owner, predecessor, and elapsed time. Almost always the bottleneck is one of: rev rec, intercompany eliminations, equity (stock comp + cap table), or FX revaluation. Pick the worst one and run a 30-day fix sprint with a named owner. Target one day shaved per sprint.

Post-close adjustments rising (count or dollars)

Play: Cutoff testing on the top five GL accounts by adjustment frequency. The pattern is usually one of three things: vendor invoices arriving after AP cutoff, revenue contracts signed at month-end without rev rec sign-off, or expense reports submitted late. Identify the top source, tighten the cutoff control, retest next close.

Control breaks > 0 at period end

Play: 30-day SLA per break with a named owner. If a break can't be closed in 30 days, it gets escalated to the CFO with a remediation plan and a hard deadline. Breaks that cross 60 days become audit findings, and your job is to never let that happen.

Audit findings aging into 90+ bucket

Play: Weekly findings-review with each finding owner. Controller chairs the meeting. Each finding gets a status, a blocker, and a date. The blocker is the only thing that matters. Solve the blocker, the finding closes itself. The Audit Committee will ask about anything in the 90+ bucket; have an answer ready before they do.

AP/AR aging spike

Play: For AP, vendor outreach within 5 days, payment plan or hard close. For AR, collections escalation matrix (CSM at day 60, AR at day 75, sales leader at day 90, legal at day 120). Tie the AR plan to the bad-debt reserve conversation; the auditor will ask if you waited too long to reserve, so document the timeline.

Accrual accuracy worsening

Play: Process audit on the worst three accruals. Replace any "last month plus vibe" calculations with a documented methodology, even if the methodology is "rolling 6-month average plus known events." A bad methodology consistently applied beats a good methodology applied inconsistently, because the auditor can defend the first one.

The Real Job

The Controller's job is not to close clean. The Controller's job is to prove the function is in control, on a trend, against a peer set.

Six metrics. Trended five quarters. Benchmarked against B2B SaaS norms. One slide. Read in 30 seconds.

That's the scorecard. Bring it to the next QBR and the conversation changes — from "how are we doing?" to "what's the plan on accrual accuracy?" The first question is unanswerable. The second one is the conversation a credible Controller wants to have.

The Controllers who get cut are the ones who answer "clean." The Controllers who get promoted are the ones who answer with a slide.

Learn More

• Controller Job Description — companion JD for the role this scorecard belongs to
• Month-End Close Checklist for Controllers
• Audit Readiness Playbook
• Controller-to-CFO Career Path